Determining the effects of League of Legends ranked tiers on outcome prediction models

League of Legends (LoL) is a popular Multiplayer Online Battle Arena (MOBA) video game with a competitive ranked system. This study trained XGBoost, Random Forest, and Logistic Regression models on Bronze and Diamond tier datasets in order to predict match outcomes as well as observe differences in model performance on Bronze to Diamond tiers. Exactly, 20,781 matches were collected using the Riot API with the intended goal of having 5000 matches for Bronze and Diamond tiers and 1500 for the remaining tiers. Datasets were created for 10, 15, 20 and 25 min for each tier. XGBoost outperformed the other models, as it achieved an accuracy of 72.7% at 10 min for both Bronze and Diamond-trained tiers. At the 15 min interval, the XGBoost model achieved an accuracy of 77.1% trained on Bronze and 76.9% trained on Diamond. At the 20 min interval, it reached an accuracy of 79.3% for the Bronze-trained model and 80.1% on the Diamond-trained model. At the 25 min interval, this accuracy increased to 82.1% for Bronze-trained and 82.8% for Diamond-trained. The Diamond test datasets between 10 to 25 min were more accurately predicted compared to other tiers.